INDEX

Explanations

instances where individuals are referred to or labelled in a specific manner

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

edia

-0.67

 conclud

-0.66

 longitudinal

-0.64

 Topics

-0.60

abal

-0.60

foreseen

-0.59

 throughput

-0.59

 sailed

-0.58

igsaw

-0.58

 trajectory

-0.58

POSITIVE LOGITS

 bluff

1.04

 liar

0.90

 unfit

0.82

 traitor

0.82

 derogatory

0.81

 irresponsible

0.76

 responsible

0.76

Champ

0.75

 coward

0.73

 insensitive

0.73

Activations Density 0.123%

instances where individuals are referred to or labelled in a specific manner

No Comments

No Known Activations